Hierarchical representation for PPI sites prediction

Table 2 Number of complexes and in each class of the Dataset

Class	Data partition	Complex	Positive		Negative
Class	Data partition	Complex	b (%)	u (%)	b (%)	u (%)
\(A_r\)	Train	8	8	9	92	91
	Validation	3	9	10	91	90
	Test	7	8	10	92	90
\(A_l\)	Train	9	14	16	86	84
	Validation	3	14	15	86	85
	Test	8	14	15	86	85
\(AB_r\)	Train	4	9	11	91	89
	Validation	3	8	9	92	91
	Test	5	8	9	92	91
\(AB_l\)	Train	4	13	13	87	87
	Validation	3	14	13	86	87
	Test	4	16	16	84	84
\(EI_r\)	Train	18	15	15	85	85
	Validation	12	15	16	85	84
	Test	14	15	16	85	84
\(EI_l\)	Train	16	29	33	71	67
	Validation	12	32	34	68	66
	Test	16	30	32	70	68
\(ER_r\)	Train	14	13	13	87	87
	Validation	3	12	12	88	88
	Test	9	11	11	89	89
\(ER_l\)	Train	10	20	21	80	79
	Validation	5	26	22	74	78
	Test	11	25	22	75	78
\(ES_r\)	Train	7	9	12	91	88
	Validation	3	10	12	90	88
	Test	7	11	12	89	88
\(ES_l\)	Train	7	25	21	75	79
	Validation	3	25	23	75	77
	Test	6	22	21	78	79
\(OG_r\)	Train	8	9	9	91	91
	Validation	3	10	9	90	91
	Test	8	12	12	88	88
\(OG_l\)	Train	9	24	24	76	76
	Validation	2	22	21	78	79
	Test	7	19	20	81	80
\(OR_r\)	Train	10	14	13	86	87
	Validation	4	12	11	88	89
	Test	9	13	14	87	86
\(OR_l\)	Train	9	23	23	77	77
	Validation	5	21	21	79	79
	Test	10	23	24	77	76
\(OX_r\)	Train	17	16	15	84	85
	Validation	11	15	15	85	85
	Test	19	14	13	86	87
\(OX_l\)	Train	16	18	18	82	82
	Validation	14	19	20	81	80
	Test	20	20	21	80	81

Positive examples are residue pairs that participate in the interface, negative examples are pairs that do not